Hi there,

after resuming from RAM (Kernel: 3.17.2, KDE 4.11.13-2 from Debian/testing), I (often) loose brightness controls from within KDE. Attempting to change brightness results in the line

Code: Select all

kded(...) PowerDevilUPowerBackend::setBrightness: org.kde.powerdevil.backlighthelper.setbrightness failed

being added to .xession-errors. Everything works as expected when first booting up. If I log out and log back into my KDE session (no restart required), everything works again.

Furthermore,

Code: Select all

echo 40 > /sys/class/backlight/acpi_video0/brightness

still works even when KDE is unable to change brightness, so the kernel does not seem to be at fault. Still, I should mention that I use https://wiki.archlinux.org/index.php/Ma ... _backlight to make screen backlight adjustments work in the first place.

I suspect there is some race condition somewhere because the behaviour described above does not always occur, i.e. sometimes I can resume just fine with screen brightness adjustment still working.

Any suggestions?

When it breaks, try killing "upowerd" then running "upower -d".
Sounds like you might be experiencing a bug in UPower.

Hi,

unfortunately, this didn't help (it also seems as if KDE normally can change brightness without upowerd running (?)).

I did notice something else, though, which might point to this being an issue elsewhere: instead of acpi_video0 I now have gmux_backlight in /sys/class/backlight, i.e. it seems there is a race condition or so in my kernel and the wrong interface shows up (?). It might not be KDE related after all.

If KDE can change the brightness without the assistance of UPower, then this points towards the graphics driver as being the source of the problem. Does logging out and back in correct the issue? (this should restart the X server, which in turn will reinitialise the graphics).

Also, could you compare the output of "lsmod" before and after suspending the system? It might also be helpful to examine the X server log as well as dmesg as they could potentially contain hints as to why the system is behaving like this.

Hi,

yep, logging out and back in does solve the problem, so it seems your analysis is right. I'll collect some more system output the next time things go wrong.